Discrete Regulated power supply

[HoracioDos]

While looking for PSU circuits with passive components I found this one.

from
https://www.homemade-circuits.com/how-to-make-versatile-variable-voltage/
It's a simplified version of a previous one. I liked this site very much.

PS: I guess it can be mirrored like this one to get positive and negative voltage.
https://www.homemade-circuits.com/0-to-50v-0-to10amp-variable-dual-power/
Is there any policy about linking content between sites?

[HoracioDos]

Here is Ltspice model for Nuno PSU. Voltage output raises very slow and there's no ripple. It will take me a while to understand how the BJT network works in both circuits and how to choose different kind of loads in ltspice. I found Kicad schematics and PCB for this one.

PS: if someone arrives to an improved and working version, please post it here.

[not1xor1]

While looking for PSU circuits with passive components I found this one.

from
https://www.homemade-circuits.com/how-to-make-versatile-variable-voltage/
It's a simplified version of a previous one. I liked this site very much.

PS: I guess it can be mirrored like this one to get positive and negative voltage.
https://www.homemade-circuits.com/0-to-50v-0-to10amp-variable-dual-power/
Is there any policy about linking content between sites?

ROTFL
"Ripple free, with less than 1Vpp" 
and 0-100V  with a 2N3055? 
and just Vbe as V reference...   

that would have been a really bad project even 50-60 years ago
now there are no excuse for such rubbish 

just forget it

[HoracioDos]

ROTFL
"Ripple free, with less than 1Vpp" 
and 0-100V  with a 2N3055? 
and just Vbe as V reference...   
that would have been a really bad project even 50-60 years ago
now there are no excuse for such rubbish 
just forget it

I see. I still don't have the skills to spot these things. Internet fooled me again, but it makes me wonder what I did wrong because I didn't see the ripple. I will check it again.

[blackdog]

Hi,

All i see is schematics who do not care about phase marging...
Like the schematic with 1000uF at the output, this must be a joke... 

Power supply's are not simpel!
Can you build a good powersupply with discrete components, yes!, but you need to follow the same rules as with a opamp power supply.

Anyone who wants to build power supply will have to learn what phase margin and gain margin is.
And do dynamic testing. :-)

Kind regard,
Blackdog

[HoracioDos]

Lol. Where can i hide? I was playing with some silly values to see if they make any difference. I should have fixed it before posting.
PS: When I read your comment for the first time about 1000uf cap on the output I thought it was the 10000uf cap in the input after the bridge rectifier. That's the one I ment. Output capacitor is not mine.....

[Cliff Matthews]

Elecktor had something basic back in '82 - should make you feel better.
http://www.retro.co.za/zs1ke/projects/PrecisionPowerSupply/PrecisionPSU-Elektor-Dec-1982.pdf

[not1xor1]

Hi,

All i see is schematics who do not care about phase marging...
Like the schematic with 1000uF at the output, this must be a joke... 

Power supply's are not simpel!
Can you build a good powersupply with discrete components, yes!, but you need to follow the same rules as with a opamp power supply.

Anyone who wants to build power supply will have to learn what phase margin and gain margin is.
And do dynamic testing. :-)

Kind regard,
Blackdog

I think the 220µF capacitor connected to the collector of Q2 is even worst than the output capacitor.

[not1xor1]

I see. I still don't have the skills to spot these things. Internet fooled me again, but it makes me wonder what I did wrong because I didn't see the ripple. I will check it again.

I guess you did not see any ripple because you were testing with no load or little current load and the 1mF output capacitor can easily take care of that.

I think the first circuit showed at the start of the thread is much better although I would change it a bit with a couple of current sources and check better the open loop response. It is probably overcompensated.
But there is no way a single 2N3055 can provide 0-38V/0-3A.
It looks like somebody just do not have any idea about power dissipation, SOA, etc..

Another good source of discrete components PSU schematics could be the service manuals of of old (50-60 years ago) bench PSUs.

[blackdog]

Hi not1xor1,

Yes, its and, and bad   
First short circuits and gone is the 2N3055 

The 220uF capacitor has a good funtion, and that is to keep the current trough R3 (Basis Q1) free of hum.
A extra resistor between C3 and Q2 would resolf a big part of the extra pole.

But... it is still bad! 

The Elektuur design from december 1982 is a bit slow but a good design to start.
It is not to difficult to make it better, but won't advise this for this beginners section.

Kind regards
Bram

[not1xor1]

Hi not1xor1,

Yes, its and, and bad   
First short circuits and gone is the 2N3055 

The 220uF capacitor has a good funtion, and that is to keep the current trough R3 (Basis Q1) free of hum.
A extra resistor between C3 and Q2 would resolf a big part of the extra pole.

But... it is still bad! 

The Elektuur design from december 1982 is a bit slow but a good design to start.
It is not to difficult to make it better, but won't advise this for this beginners section.

Kind regards
Bram

I do not agree about the 220µF capacitor. It affects the transient response much more than the output capacitor.

The right way to reduce the ripple in that kind of circuit is to use two resistors instead of just one (i.e. R1).
For instance a 1.5k resistor plus the 220µF capacitor would reduce by a fair amount the ripple, while another 1.8k resistor would improve the transient response.

Besides that ugly circuit, the Elektor circuit diagram, is affected by the same problem. Even just 4.7µF is enough to produce a really ugly transient response.
That capacitor was probably added as an afterthought because that circuit oscillates badly without it.
It is just a matter of ignorance. The BJTs composing that sort of sziklay configuration are badly polarized and do not work at their best.

Just a couple of resistors between the base and emitter of T1 and T2 make the circuit behave properly and the 4.7µF (C2) capacitor is no longer necessary to avoid oscillations while the transient response is greatly improved.

There are other bad design choices and other possible improvements in that Elektor circuit, but I'll leave that to another future post, if anybody is interested.


Elektor schematic LTspice circuit simulation (current limit removed)


Output ripple at 2.5A load


The ugly transient response from 100mA up to 2.5A


The current through the 3 output BJTs


Better BJTs polarization and a little improvement in transient response


Modification of the compensation network and better transient response

[HoracioDos]

It's very interesting to see how analysis and improvement is done.
Could you please tell transient and load parameters.
Thank you very much.

There are other bad design choices and other possible improvements in that Elektor circuit, but I'll leave that to another future post, if anybody is interested.

I'm interested but I don't want to take advantage of your generosity

[techguru]

thank you  not1xor1 sir,

                      As a student give me the path by which you have acquired this knowledge. No text book worries about gain and phase margin.

[not1xor1]

It's very interesting to see how analysis and improvement is done.
Could you please tell transient and load parameters.
Thank you very much.

There are other bad design choices and other possible improvements in that Elektor circuit, but I'll leave that to another future post, if anybody is interested.

I'm interested but I don't want to take advantage of your generosity

I do not have much time at the moment.
Anyway I found a good solution to make a simple yet better current limitation circuit. I also replaced the two zeners with a TL431.

According to LTspice the circuit even behaves nicely at turn-on and turn-off with the output voltage increasing/decreasing slowly.

In a couple of days I hope I'll be able to post a zip with the various models and test circuits.

[blackdog]

Hi,

No Spice for me, just solder. 

I hackt the schematic and below you can see the result for now.


Selecting and measuring the transitors with a cheap China meter, handy!.



Testing the current source and the zener with a variable power supply to see how mutch the zener voltage wil change.



I am now about half of building the circuit, the transistors 1 and 2 in the schematic are tight together for better thermal behavior.



I have adapted this version so that I can supply 22V and 2-Ampere.
If the circuit works, I will test the dynamic behaviour of this circuit.
The reason for using a modern power transistor is that it will then be easier to keep the circuit stable.
I have to confess that I have never built a Power Supply with a Darlington that has three transitors.
Probably there will be another HF stop resistor around this transitor.

Because I would like to have a reasonable noise / hum free design, the number of parts used is increasing rapidly.
And also three extra diodes for safety.

But be aware that with two opamps and a TL431 you can make a better power supply! 
Why then all that effort to even build it, because I like to keep everything discreet and also just to play with it...
Depending on my work in the coming week, I will show the outcome of the above schematic.

Kind regards,
Bram

[blackdog]

Hi!


Yes i had some spare time today, to hot to go outside here in Amsterdam 

Schematic update:



And this is the Power Section, build on a CPU heatsink, everything short wired.



Almost done, only the ground and de +bar on the bottum side.



Not yet cleaned and missing two wires like i expland, row-01 (ground) and row-18 (+)



Shoot @ It!

Regards,
Bram

[Cliff Matthews]

@Blackdog - I've seen 2SC5200 in audio amps.. OK you twisted my arm  Digikey has them at $2 ea. if I buy 10.

Thanks, I am looking forward to see your perfections on this design sir - great work as usual!

[blackdog]

Hi Cliff,

Do not order or buy anything, let me first do some tests...

Circuitboard cleant with IPA and a old toothbrush, i hope is stays this way 


Heathsink mounted with a Ty-Wrap.



Frontview of the testsetup.



Backside.



And now its time to doe some work!
Later more.

Kind regards,
Bram

[Cliff Matthews]

Looking good!  (Q6 needs a label correction though, details details..)

[HoracioDos]

For those who might be interested. I hope I didn't make any mistake while copying the schematic.

PS: I've fixed the schematic around 6.2V zener as it will be seen next

[blackdog]

Hi HoracioDos, 

Of course you made at least 1 mistake! 
Back to the drawing board, the parts around the 6.2V zener are wrongly drawn.

Thank you for showing this spice setup!

Kind regards,
Bram

[HoracioDos]

Of course you made at least 1 mistake! 
Back to the drawing board, the parts around the 6.2V zener are wrongly drawn.

Thanks Bram! I'll try to fix it and I'll upload it again in the same post.
PS: Done!

[HoracioDos]

Hello.
This is the model for the elektor 1982 PSU. Simulation lasts forever. I can't find the problem (I'm pretty tired right now). I wanted to see if the LM741 can be changed to a TL071. Model and the LM723 part are inside the zip file. I'll be grateful if someone makes it work.

[not1xor1]

Hello.
This is the model for the elektor 1982 PSU. Simulation lasts forever. I can't find the problem (I'm pretty tired right now). I wanted to see if the LM741 can be changed to a TL071. Model and the LM723 part are inside the zip file. I'll be grateful if someone makes it work.

that is the classical Harrison schematic:
http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1962-07.pdf
The LM723 is used just as voltage reference. You might use any other reference in the simulation.

BTW you can find many useful informations about power supplies in this really interesting application note of HP/Agilent
http://literature.cdn.keysight.com/litweb/pdf/5952-4020.pdf

[Cliff Matthews]

Hi Cliff,

Do not order or buy anything, let me first do some tests...

Circuitboard cleant with IPA and a old toothbrush, i hope is stays this way 


Heathsink mounted with a Ty-Wrap.



Frontview of the testsetup.



Backside.



And now its time to doe some work!
Later more.

Kind regards,
Bram

Will we be privileged to see your finished product? (and maybe any final schematic/asc corrections?)